Come on now, are we really using fizzbuzz as a performance benchmark? Seriously, don't base a real-world decisions on this.

PS. The Rust implementation is also not very optimal, for example, the stdout is not locked so... what are we even trying to show :-).